[ndnSIM] Counting the number of interest packets received in a Namespace

[Dyaneswaran S]

Hi guys,
    My objective is to count the no of interest packets received in a node.
Since this looked like the kind of namespace specific information that can
be stored in the Measurements Table, I did the following as suggested in
the NFD Developers Guide and other strategies in the forwarder,

1. Created a Nested Counter Class inside my Strategy class. It looks
something like this -

  class CounterInfo : public StrategyInfo
  {
        public:
        static constexpr int
        getTypeId()
        {   return 1009; }

        CounterInfo() : m_count(0)
        {
        }

        void
        increment()
        {  ++m_count;}

        uint64_t
        get()
        {   return m_count; }
        private:
        uint64_t m_count;
  };

2. Then, in my AfterReceiveInterest() Trigger, I retrieve the entry using
the piEntry APIs like this,

          CounterInfo *counter = pitEntry->getStrategyInfo<CounterInfo>();
          if(pitEntry->getStrategyInfo<CounterInfo>() == nullptr)
          {
                counter = new CounterInfo();
                counter->increment();
                pitEntry->insertStrategyInfo<CounterInfo>(*counter);
          }
          else
          {
                counter->increment();
          }

3. I thought the above 2 steps were sufficient, but apparently not. So
after more digging, I found out I had to extend the Lifetime of
Measurements Table Entries. I was not sure how Measurements and
StrategyInfo Class were interrelated but I did it anyway. (superstitious, I
agree).  That looked like -

 this->getMeasurements().extendLifetime(*entry, MEASUREMENTS_LIFETIME);

But even then, I didnt get the expected output. That is the counter
variable is not counting the packets.

Instead, its count never gets past 1. I am sure this is because its getting
instantiated newly everytime this strategy is applied.

Somehow, its not accessing the already stored strategy info. I thought it
was because of the different pit entries causing new measurement entries
(since measurements are accessed via pitentries). But I made sure that
didnt happen as well.

I thought about approximating the count of interest packets to the no. of
PIT inrecords but that was ridiculous. I needed the actual count of
interest packets.

Any pointers?? Any of my assumptions wrong?
Am I doing it the hard/correct way? Any other methods?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20200205/04426c9a/attachment.html>